BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20250626T234541Z
LOCATION:B304
DTSTART;TZID=America/New_York:20241117T140000
DTEND;TZID=America/New_York:20241117T144500
UID:submissions.supercomputing.org_SC24_sess745_misc208@linklings.com
SUMMARY:Invited Talk: Mapping Irregular Computations to Accelerator-Based 
 Exascale Systems
DESCRIPTION:Katherine Yelick (University of California, Berkeley; Lawrence
  Berkeley National Laboratory (LBNL))\n\nAs traditional technology drivers
  of computing performance level off, the use of accelerators with various 
 levels of specialization, are growing in importance. At the same time, dat
 a movement continues to dominate running time and energy costs, making com
 munication cost reduction the primary optimization criteria for compilers 
 and programmers.  This requires new ways of thinking about algorithms to m
 inimize and hide communication, expose fine-grained parallelism, and manag
 e communication. These changes will affect the theoretical models of compu
 ting, the analysis of performance, the design of algorithms, and the pract
 ice of programming.  \n\nIn this talk I will discuss prior work and open p
 roblems in optimizing communication, avoiding synchronization, and tolerat
 ing nondeterminism, using data analysis and statistical learning problems 
 from biology as driving examples. I will discuss distributed data structur
 es and communication optimizations in large-scale genome analysis, includi
 ng metagenome assembly, protein clustering, and more. The algorithms repre
 sented data analysis “motifs” including hashing, alignment, generalized n-
 body, and sparse matrices.  I will describe two parallelization approaches
 , one based on asynchronous one-sided communication and another based on b
 ulk-synchronous collectives using GraphBLAS. I will give an overview of th
 ese approaches, describe the GPU parallelizations, and highlight some of t
 he resulting scientific insights, including the discovery of new microbial
  species and new protein functional dark matter.\n\nTag: Algorithms, Heter
 ogeneous Computing\n\nRegistration Category: Workshop Reg Pass\n\nSession 
 Chairs: Vassil Alexandrov (Hartree Centre); Jack Dongarra (University of T
 ennessee, Oak Ridge National Laboratory (ORNL)); Erik Draeger (Lawrence Li
 vermore National Laboratory (LLNL), Center for Applied Scientific Computin
 g); Christian Engelmann (Oak Ridge National Laboratory (ORNL)); and Dieter
  A. Kranzlmueller (Ludwig-Maxmilians-Universität München, Leibniz Supercom
 puting Centre (LRZ))\n\n
END:VEVENT
END:VCALENDAR